The common constraint in the research on post-disaster road network repair decision problems is that each damaged road segment is repaired by a single crew, which limits the possibility of reducing the repair time of the important road segment through multi-crew cooperation, so as to improve the repair effect. In order to solve this problem, this paper proposes the constraints of multi-crew cooperation based on the damaged road network model and the synchronization of multi-crew, and studies the multi-crew coordinated scheduling and routing problem in the post-disaster emergency rescue stage with the goal of maximizing the cumulative utility of road network performance. The road network performance index is established according to node accessibility after disaster, and the reward function of repair action under the multi-crew scenario is designed. The Markov decision process of the multi-crew coordinated scheduling and routing problem is constructed, and the greedy algorithm and Q learning algorithm are used to solve the problem. Finally, the case analysis shows that the proposed method can accelerate the repair of road network performance, effectively improve the cumulative utility of road network performance in the emergency rescue stage, and ensure the smooth progress of post-disaster rescue activities, no matter in the random failure scenario or disaster point failure scenario.